The Impact of Instrument Fracture on Outcome ofEndodontic TreatmentPeter Spili, BSc, BDSc, Peter Parashos, MDSc, PhD, and Harold H. Messer, MDSc, PhD

AbstractAdoption of rotary nickel-titanium instruments has re-newed concerns regarding instrument fracture and itsconsequences. The frequency of instrument fractureand its impact on treatment outcome were determinedfrom an analysis of specialist endodontic practicerecords involving 8460 cases. A case-control study oftreatment outcomes was conducted on a subset of 146teeth with a retained instrument fragment (plus 146matched controls), for which clinical and radiographicfollow-up of at least 1 year was available. Maskedradiographs were assessed by two calibrated examin-ers. Overall prevalence of retained fractured instru-ments was 3.3% of treated teeth. In the case-controlstudy, overall healing rates were 91.8% for cases witha fractured instrument and 94.5% for matched controls(p � 0.05, Fisher’s exact test). Healing in both groupswas lower in teeth with a preoperative periapical radi-olucency (86.7% versus 92.9%, p � 0.05). In the handsof skilled endodontists prognosis was not significantlyaffected by the presence of a retained fractured instru-ment.

Key WordsTreatment outcome, prognosis, fractured instruments,fractured instrument prevalence, case-control study

From the School of Dental Science, University of Mel-bourne, Melbourne, Victoria, Australia.

Address requests for reprint to Professor Harold H. Messer,School of Dental Science, University of Melbourne, 711 Eliza-beth Street, Melbourne, Victoria 3000, Australia. E-mail ad-dress: hhm@unimelb.edu.au.

Copyright © 2005 by the American Association ofEndodontists

Only a few small studies have evaluated the effect of a retained fractured endodonticinstrument on the outcome of endodontic treatment (summarized in Table 1).

Those that have been regularly cited include two early studies, which reported that theretained fragment reduced healing, particularly in the presence of a preexisting peria-pical radiolucency (1, 2), and others, which stated that it had no influence on healing(3, 4). Some of these studies further suggested that in most cases the retained fracturedinstrument can be incorporated into the final root canal filling (4, 5), although recentlydeveloped techniques have made removal more predictable (6 – 8). The conflict ofopinion regarding the clinical significance of retained fractured instruments makes itimperative that clinicians be offered more definitive, evidence-based information forpredicting the potential consequences of this procedural complication.

A brief summary of the generally recognized hierarchy in levels of evidence inclinical studies is, in descending order, randomized controlled trials, cohort studies,case-control studies, case series, and case reports (9, 10). To date, the availableevidence regarding prognosis of retained fractured instruments has come from studiesthat have been mostly case series, which offer a low level of evidence, and only one truecase-control investigation (4). Retrospective case-control studies realistically providethe highest level of evidence possible in such investigations. An important finding frommany of the studies that evaluated the prognostic impact of a retained instrumentfragment was that the presence of a preoperative periapical lesion served as the mainprognostic factor for the successful treatment of such cases (1, 2, 5, 11, 12).

Fractured root canal instruments may include endodontic files, lateral or fingerspreaders, spiral fillers, or Gates-Glidden burs, and can be nickel-titanium (NiTi),stainless steel (SS), or carbon steel. The prevalence of retained fractured endodontichand instruments (mostly SS files) has been reported to range from 1 to 6% (1, 4, 11,13–17). The fracture incidence of rotary NiTi files, based on only a small number of veryrecent investigations, has been shown to lie in the range 0.4 to 5% (17–19); the higherfigure represented fractures only in molar teeth (18). In most circumstances fractureresults from incorrect use or overuse of an endodontic instrument, and occurs mostcommonly in the apical third of a root canal (20). Although there is a perception thatrotary NiTi instruments may fracture without warning, recent work indicates that frac-ture involves many factors the most important of which seems to be the clinician’s skill(21).

Since the introduction of NiTi instrumentation over 10 yr ago, no published clinicaloutcome study has reported the influence of fractured rotary NiTi instruments on theprognosis of endodontic treatment and retreatment. Therefore, the purposes of thisstudy were: (a) to determine the frequency of fractured endodontic instruments left inthe root canal following treatment, and (b) to assess the influence of endodonticinstrument fracture, in particular rotary NiTi, on the outcome of nonsurgical root canaltreatment and retreatment, in specialist endodontic practice. The null hypothesis testedin the second part of the study was that a retained instrument fragment does notinfluence postoperative healing.

Materials and MethodsStudy Design

This study was divided into two parts: 1) a retrospective survey of all cases treatednonsurgically (including retreatment) over some 13.5 yr in two endodontic practicesinvolving seven endodontists. The main objective was to determine the prevalence ofretained fractured instruments following treatment; 2) the second part was based on all

Clinical Research

JOE — Volume 31, Number 12, December 2005 Fractured Instruments and Treatment Outcome 845

cases from part 1 for which clinical and radiographic follow-up for atleast 1 yr was available. Specifically, it entailed a retrospective, case-control treatment outcome analysis of equivalently matched cases,taken from the pool of cases of part 1, differing only by the presence orabsence of a retained fractured endodontic instrument.

Ethics approval for the project was obtained from the University ofMelbourne Human Research Ethics Committee. All patient records in-cluded for analysis were de-identified through a suitable code that main-tained their anonymity throughout the study.

Part 1: Frequency of Instrument FractureThe overall prevalence of all fractured instruments retained in the

root canal was determined from scrutiny of the records and radio-graphs of 8460 cases treated by seven endodontists in two private prac-tices in Melbourne, Australia, during the period January 1990 to May2003. Because rotary NiTi instrumentation was adopted as the primarytechnique for canal preparation in the two practices in 1996 to 1997,data from 1997 onwards were analyzed separately, for five endodon-tists, to determine the incidence only of retained rotary NiTi and SS handfiles (used only as pathfinders) from 5103 cases out of the total pool of8460 cases. The major rotary NiTi instruments used during that periodincluded: Quantec Series 2000 (NT Co., Chattanooga, TN); Quantec LX& SC (Analytic Endodontics, Glendora, CA); GT Rotary (Dentsply-Tulsa

Dental Products, Tulsa, OK); ProFile 0.04 and 0.06 taper and ProTaper(Dentsply Maillefer, Ballaigues, Switzerland); and K3 (SybronEndo, Or-ange, CA). All instruments were used according to the manufacturers’recommendations. In general, multivisit endodontics was performedwith calcium hydroxide paste (Pulpdent Corp, Watertown, MA) and/orLedermix paste (Lederle Pharmaceuticals, Wolfratshausen, Germany)used as the interappointment medicament(s). Obturation in most casesinvolved cold lateral compaction using gutta-percha and AH26 sealercement (Dentsply De Trey GmbH, Konstanz, Germany).

Part 2: Case-Control StudyDuring the initial screening of patient records, all 8460 cases

treated with orthograde endodontic treatment or retreatment during the13.5 yr period were coded for the following potential outcome-influ-encing variables:

• Patient age• Patient gender• Tooth type• Presence or absence of periradicular/periapical lesion• Nonsurgical primary treatment or retreatment• Recall period, coded as: 1 to 2 yr, 2 to 4 yr, or �4 yr

TABLE 1. Studies reporting the effect of a retained fractured instrumenta on the outcome of endodontic treatment

Study Observationperiod (yr)

Controlgroup

Preoperative periapical status Outcome Stated effect offractured instrument

on healingLesion No lesion Success (%)

Strindberg (1) 4–10 Yes, but notcase-control

2/4b 9/11 11/15 (73%)c Overall 19% reductionc

Grahnén & Hansson(25)

4–5 No NR NR NS No effect

Ingle & Glick(3) 0.5–5 No NR NR NS No effectEngström et al (13) 4–5 Nod NR NR 6/9 (67%) No effectEngström &

Lundberg (14)3.5–4 Nod 0/0e 5/5 5/5 (100%) No effect

Grossman (2) 0.5–6(average 2)

Nof 9/19 42/47g 51/66 (77%)g Reduced only whenlesion present

Crump & Natkin (4) �2 Yes,case-control

27/29 21/24 48/53 (91%) No effecth

Fox et al (5) 2–14(average 2–3)

No NR NR 93/100 (93%)i Reduced only whenlesion present

Bergenholtz et al(11)j

2 No NR NR NS Reduced only whenlesion present

Kerekes & Tronstad(15)

3–5 No NR NR 9/11 (82%)k Reduced only in teethwith necrotic pulps

Cvek et al (31) 4 No 3/4l NAl 3/4 (75%) Not stated specificallyfor fractured files

Sjögren et al (16) 8–10 No NR NR 9/11 (82%)k NSMolyvdas et al (12) 0.5–5

(average 3)No 8/11m 32/35m 40/46 (87%)m Reduced only when

lesion presentTotal (%) 49/67 (73%) 109/122 (89%) 275/320 (86%)

NR � Not reported, NS � Not stated.a All instruments were either carbon steel or stainless steel hand files.b Number of cases judged to be successful over total number of cases.c At final observation. Also, this author did not classify ‘incomplete healing’ as success.d Comparison between cases based only on presence or absence of positive bacterial culture before obturation.e All cases in this study were treated only with pulpectomy, hence no cases with lesions.f Group of ‘normal’ cases, not clearly defined as a control group, were used for comparison.g If ‘doubtful’ cases are considered failures.h No statistically significant difference between cases and controls.i Only the cases with ‘accidentally’ fractured files are considered in this table.j Retreatment cases only.k Roots, as opposed to teeth in other studies.l All cases in this study had posttraumatically reduced pulpal lumens and preoperative periapical lesions.m If ‘compromised’ cases are considered failures.

Clinical Research

846 Spili et al. JOE — Volume 31, Number 12, December 2005

• Month and year of root filling• Operator

Additional variables coded for the fractured instrument cases in-cluded:

• Type of instrument: rotary NiTi, SS hand file, paste filler• Root or canal in which the fracture occurred• Level in the root canal where the fracture occurred: coronal third,

middle third, apical third, apical � beyond apex• If the fractured instrument was bypassed or not

From this large pool of screened records, all fractured instrumentcases and nonfractured instrument controls for which a minimum of1-yr clinical and radiographic follow-up was available were identified.Each fractured instrument case was then matched as closely as possiblewith an equivalent nonfractured instrument control for the factors listedabove. The appropriately matching control for each case was randomlyselected from the pool of potential control teeth that comprised 812primary treatment and 153 retreatment. All relevant clinical and radio-graphic data were coded and entered in Excel spreadsheets (MicrosoftCorporation, Redmond, WA).

Exclusion Criteria for Both Fractured Instrument and ControlCases

• Any previously root-filled tooth already containing a fractured instru-ment

• Any nonsurgically treated or retreated tooth with a clearly defectivecoronal restoration or obvious coronal leakage that was visible onthe radiograph or noted in the record

• Any case with unsatisfactory or missing radiographs and/or one thatwas poorly documented

Radiographic ExaminationFor each fractured instrument case and its matched control, a

preoperative, postobturation, and appropriate recall radiograph, weresequentially mounted on gray cardboard radiographic film mounts(Rinn, Dentsply, Surrey, UK). Each film mount was first covered with aclear transparency plastic sheet (Dataline plain transparency film, Es-selte, EU). Then, on this sheet, every root containing a fractured instru-ment was masked with a dense black permanent marker that completelyconcealed any underlying root filling material and fractured instrument,but making sure all periapical structures of the masked root, includingall other roots for multi-rooted teeth, were still visible for assessment.All equivalent roots of control cases were masked in exactly the samemanner. Each radiographic series was given a numerical code andrandomly distributed in preparation for rating.

Two independent endodontists were calibrated (see below) andsubsequently blindly evaluated all coded radiographs. Viewing was per-formed in a darkened room using a well-illuminated 38 cm � 16 cmviewbox, with complete masking of peripheral light, and a magnifierviewer with 2� magnification (X-Produkter, Malmo, Sweden). The unitof radiographic evaluation was (a) the whole tooth, and (b) the indi-vidual root. The unit of clinical evaluation was the whole tooth.

Calibration of ExaminersEach examiner was individually calibrated, for the presence or

absence of a periapical lesion, by rating the preoperative and finalradiographs of a random sample of 60 cases. Cases were also rated forthe following four radiographic categories of healing:

a) Complete healing (C), total resolution or no development oflesion on recall radiograph; b) Incomplete healing (I), radiographic

signs of obvious, but incomplete, bone repair, including repair aroundexcess root filling material; c) Uncertain (U), unclear whether therehad been any change in the lesion on the recall radiograph, but theclinical record did not indicate any clinical signs or symptoms; d) Nohealing (F), lesion on recall radiograph was clearly either enlarged ordefinitely unchanged, or a new lesion had developed. For multirootedteeth, the overall healing category assigned to the tooth was based on theroot exhibiting the worst outcome.

After a 2-wk interval, the two examiners re-rated the same set ofradiographs. Cohen’s Kappa values for inter- and intraobserver reliabil-ity were then calculated. When they evaluated each of the 292 cases(146 case teeth and 146 control teeth) for the main part of the study, anydisagreement on the final rating was resolved by reaching a consensusbetween the two examiners. If they still disagreed on a particular case,the opinion of the principal author (PS), who was also blinded to thecase being evaluated, was accepted as final. The two examiners’ inter-pretations of presence or absence of a preoperative lesion differedslightly from the principal author’s interpretation during his initialscreening of cases and selection of matched pairs. Therefore, this ac-counted for the small discrepancy of one case each in the number ofteeth with or without a lesion between the matched pairs in Table 3 ofthe Results.

Outcome AssessmentBoth radiographic and clinical criteria were used to assess out-

come. Success was defined in terms of “healing” and encompassed: (a)radiographic categories of complete healing and incomplete healing,and (b) absence of clinical signs and symptoms. Failure encompassed:(a) radiographic categories of uncertain and no healing, and/or (b)presence of clinical signs and symptoms.

Recall PeriodCases were allocated to one of the following three categories: a) 1

to 2 yr, b) 2 to 4 yr, or c) greater than 4 yr.

Statistical AnalysisAll statistical analyses were performed in Minitab version 14.11

(Minitab Inc., State College, PA). Comparison of treatment outcome,and the effect of preoperative periapical status on outcome, betweenfractured instrument cases and controls was determined using Fisher’sexact test. The same comparisons were made between fractured rotaryNiTi files and SS hand files also using Fisher’s exact test. The dependentvariable for all analyses was outcome, which was collapsed into twocategories: a) Success � complete or incomplete healing, and b) Fail-ure � uncertain or no healing/failure. Stepwise logistic regressionanalysis was performed to evaluate associations among various poten-tial outcome-influencing variables and treatment outcome. The level ofsignificance was set at 0.05 for all analyses.

ResultsCohen’s Kappa Values

The Kappa value for the interobserver agreement was � � 0.73.The intraobserver Kappa values for the two raters were � � 0.68 and0.78. All three Kappa values were in the “substantial agreement” cate-gory (22).

Fracture PrevalenceOverall, 277 teeth from 8460 teeth were identified as containing

one or more instrument fragments (total � 301 fragments), yielding aprevalence of all types of retained fractured endodontic instruments of3.3%, from two specialist practices screened for the period January1990 to May 2003 (Table 2). Of the instrument types, rotary NiTi in-

Clinical Research

JOE — Volume 31, Number 12, December 2005 Fractured Instruments and Treatment Outcome 847

struments accounted for 78.1% of the total, SS 15.9%, paste filler 4.0%,and lateral spreader 2.0%. The 277 teeth with fractured instrumentswere identified from 263 patients (118 males and 145 females). Ofthese 263 patients, 143 (average age range 40 – 60 yr) with 146 teeth,had presented for recall of �1 yr.

During the period January 1997 to May 2003 (n � 5,103cases) the overall prevalence of retained fractured instruments (ro-tary NiTi, n � 226; SS hand file, n � 35), for the five endodontistswas 5.1% (range: 3.5–5.3%). There was no statistically significantdifference in fracture prevalence, and no significant change in frac-ture rate over time; specifically, there was no consistent pattern ofdeclining fracture rate with increasing years of experience amongthe five endodontists. Of the overall fracture prevalence of 5.1%,4.4% were NiTi and 0.7% were SS.

Case-Control Study of Impact on Treatment OutcomeOf the 146 teeth with �1-yr recall, 158 roots were found to con-

tain fractured instruments. The main unit of analysis chosen for thisstudy was “teeth” rather than “roots.” When outcomes were determinedon an individual root basis, they differed from the data for whole teeth byno more than � 1.5%. The overall success rate (C�I) of cases andcontrols was 93.7% (Table 3), specifically, 91.8% for cases and 94.5%for controls. This 2.7% difference was not statistically significant (p �0.49, Fisher’s exact test). Overall, teeth without a preoperative periapi-cal lesion showed a higher success rate than teeth with a lesion present.In the absence of a periapical lesion, the presence of a fractured instru-ment had no effect on outcome (98.4% versus 96.8%[control], p �0.62, Fisher’s exact test). When the tooth had an associated periapicallesion, healing was lower when a fractured instrument was present(86.7% versus 92.9%[control]), but this 6.2% difference was not sta-tistically significant (95% confidence interval: �3.0% to 15.3%; p �0.21, Fisher’s exact test).

The stepwise logistic regression analysis revealed that only thepresence of a preoperative periapical lesion was significantly associatedwith a reduced chance of healing, after adjustment for the other vari-ables (p � 0.015). The odds of a successful outcome were estimated tobe 4.8 times greater in the absence than in the presence of a lesion.Other variables such as gender, age, tooth type, primary or retreatment,recall period, level of instrument fracture in the root canal, and whetheror not the fractured instrument was bypassed did not significantly affectoutcome (data not shown).

A comparison of the effect of fractured rotary NiTi files versus SShand files on overall outcome (Table 4) revealed 113 of 119 (95.0%)rotary NiTi cases were successes, and 30 of 33 (90.9%) SS hand filecases were successes. This difference was not statistically significant(p � 0.41, Fisher’s exact test). There was no statistically significantinfluence on these data by the presence or absence of a preoperativeperiapical lesion (Table 4). When looking at the time-course of healingafter treatment, the number of incomplete healing cases decreasedmarkedly with increasing recall period (Table 5), and this was signifi-cant for both fractured instrument cases and controls. However, per-cent healed (as defined by C�I) was not affected by the recall period forthe control cases.

DiscussionThe prospective, randomized controlled clinical trial is the gold

standard by which all clinical research is judged because it representsthe highest grade of evidence that can be used for the provision ofevidence-based health care (9). However, the best level of evidence thatcan be realistically achieved for an evaluation of the impact of a proce-dural complication such as a fractured instrument on endodontic treat-ment outcome, is a case-control study; although, some would argue thata well-designed observational study (such as one with a case-controldesign) can produce results similar to that of a randomized controlledtrial (10). Retrospective case-control studies are difficult to performbecause it is challenging to identify a large number of cases and then toidentify appropriate controls based on multiple selection criteria. Oursample of 146 pairs was a large sample size for a study of this kind, andwas comparable with all previous studies combined (Table 1).

TABLE 2. Descriptive data for all fractured instrument cases involving 301retained fragments in 277 teeth (January 1990-May 2003)

Total (%n)

Tooth type Mx anterior 11 (4.0%)Md anterior 2 (0.7%)Mx premolar 19 (6.9%)Md premolar 9 (3.2%)Mx molar 110 (39.7%)Md molar 126 (45.5%)n 277

Treatment Primary 254 (91.7%)Retreatment 23 (8.3%)n 277

Periapical status Lesion 153 (55.2%)No lesion 124 (44.8%)n 277

Instrument type Rotary NiTi* 235 (78.1%)SS hand file 48 (15.9%)Paste filler 12 (4.0%)Lateralspreader†

6 (2.0%)

n** 301Location of fragment Coronal 1⁄3 1 (0.3%)

Middle 1⁄3 57 (18.9%)Apical 1⁄3 232 (77.1%)Beyond apex 11 (3.7%)n 301

*Rotary NiTi became the primary canal preparation technique from 1997 onwards.

**Some of the 277 teeth with fractured instruments contained more than one fractured instrument,

hence the higher number of instruments than teeth.

†The fractured lateral spreader tips were not included in the case-control study because they were not

considered to affect outcome at the stage when they were fractured.

TABLE 3. Outcome of case-control study according to preoperative periapicalstatus (teeth)

No. ofteeth (n) C I U F % healed

(C�I)

Fractured instrumentNo lesion 63 62 0 0 1 98.4Lesion 83 50 22 1 10 86.7Case-controlNo lesion 62 60 0 0 2 96.8Lesion 84 45 33 0 6 92.9Total 292 217 55 1 19 93.7

C � complete healing, I � incomplete healing, U � uncertain, F � no healing.

TABLE 4. Relationship between fractured instrument type and preoperativeperiapical status on healing (roots)

No lesion Lesion

Success (%) Success (%) Total success (%)NiTi 62/63* (98.4) 51/56 (91.1) 113/119 (95.0)SS 16/16 (100) 14/17 (82.4) 30/33 (90.9)Paste filler 2/2 (100) 3/4 (75.0) 5/6 (83.3)Total 80/81 (99.0) 68/77 (88.3) 148/158 (93.7)

*Number of instruments judged to be successful over total number of instruments.

Clinical Research

848 Spili et al. JOE — Volume 31, Number 12, December 2005

This retrospective, clinical and radiographic evaluation of frac-tured-instrument cases versus matched controls revealed that the over-all success rate of endodontic treatment and retreatment when a frac-tured instrument was left within a root canal (91.8%) was notsignificantly different from that of the controls (94.5%). When the datawere analyzed according to the absence or presence of a preoperativeperiapical lesion (Table 3), again no statistically significant effect of afractured instrument was found. Overall, our success rates were veryhigh, but within the range of previously reported studies for overallsuccess and failure of endodontic treatment (23). Our findings supportthose of Crump and Natkin (4) who performed the only comparablecase-control study, but which involved mainly silver cone obturation.From their sample of 53 matched pairs, they also showed that there wasno statistically significant difference in failure rate between broken in-strument cases and controls. Additionally, based on the period whentheir cases were completed (1955–1965), instruments were either car-bon steel or ss hand instruments. One other frequently cited controlled(but not case-controlled) study that also looked at this issue was de-scribed by Strindberg (1). He reported a significantly lower (19%) rateof healing compared to a much larger “control” group; however, thisfinding was based on only 15 fractured-instrument cases of which fourhad periapical lesions, and the author considered incomplete healing as“failure.” In contrast to all previous reports, our study included pre-dominantly rotary NiTi instruments, and showed that retained rotaryNiTi fragments do not adversely affect outcome any more than do steelhand files. Although we found only a small effect on outcome for aretained fractured instrument, it is likely that the ultimate prognosis maydepend on the stage and degree of canal preparation when instrumentbreakage occurs and, therefore, the extent to which microbial control iscompromised (1, 4, 12, 24). This information was not available fromour study sample.

When endodontic treatment is performed to a high technical stan-dard, the influence of a periapical lesion appears to be slight, but if thetechnical standard is compromised, the presence of a lesion can reducethe success rate considerably (1, 13, 16, 25). Hence it might be pre-dicted that a fractured instrument would reduce the prognosis of teethwith a periapical lesion, based on the fact that the technical standard islikely to be compromised when the fractured instrument is present.Within the fractured instrument cases (Table 3), those with lesions were11.7% less likely to heal than those without a lesion, which was statis-tically significant. However, the presence of a fractured instrument initself did not compromise the outcome. This was confirmed statisticallyusing logistic regression analysis, which accounts simultaneously forseveral explanatory variables—in this case, outcome-influencing vari-ables pertaining to endodontic treatment. Of all variables examined,only presence of a periapical lesion had a statistically significant influ-

ence on healing. In teeth with a periapical lesion, a 6.2% lower rate ofhealing was found when a fractured instrument was present, which wasnot statistically significant. The wide confidence interval of this differ-ence (�3.0 to 15.3%) makes it difficult to assign any clinical signifi-cance to this result.

This study found an overall fracture prevalence of 3.3% for alltypes of retained fractured instruments in the root canals of the totalstudy sample. During the period when rotary NiTi was the primarytechnique for canal preparation (1997 onwards), the percentage ofteeth with retained fractured rotary NiTi files was 4.4%. This falls withinthe range of 0.4 to 5% reported in previous clinical studies for rotaryNiTi files (17, 18), and exceeds that of fractured SS hand files (0.7%).In direct contrast, Pettiette et al. (17) reported a fracture incidence of0.4% for rotary NiTi and 1.1% for SS files. However, their sample wasmuch smaller and the operators were undergraduate students whoprobably treated much simpler cases. In our study, as with the study byPettiette et al. (17), the SS files were generally used only to establishpatency and provide a glide path for the rotary instruments. Further,only the very small sizes (#06-15) were generally used and they werealmost exclusively new instruments for each case. The rotary NiTi files,on the other hand, were used up to five times and considerably morewere used per case than the SS hand files.

The yearly fracture rates for all endodontists did not show anyconsistent pattern of decline with increasing time or operator experi-ence. A possible explanation for this finding is that between 1997 and2003 several new designs of rotary NiTi instruments were introducedonto the market, which meant that each had its own associated learningcurve. Consequently, the clinicians had to adapt to the new character-istics of the new instruments. Operator-related factors have a significantinfluence on fracture incidence of rotary NiTi instruments (21).

As with other retrospective outcome studies, a number of limita-tions must be recognized because not all outcome-influencing variablescan necessarily be controlled. One of these is the recall period. In thisstudy, 56.8% of all cases had an observation period of 1 to 2 yr, with21.7% of these being incomplete healing, which are likely to lead tocomplete healing (1, 15, 26 –28, 32). Additionally, mainly nonstand-ardized radiographs were available, but were acceptable for this studybecause it was a retrospective field study, and reliability of scorers wasassessed. The “substantial agreement” kappa values reflect favorably onthe results of this study and the means of calibration. Because mostradiographs were nonstandardized, the PAI scoring system (29, 30)could not be used.

It appears from our study that a fractured rotary NiTi instrumentretained in a root canal does not impact adversely on prognosis, pos-sibly because treatment was performed to a high technical standard,including strict aseptic techniques, by experienced endodontic opera-tors. Whether these findings can be extrapolated to general practiceconditions remains to be determined.

Conclusionsa) The overall frequency of fractured endodontic instruments left

in the root canal after treatment was found to be 3.3% of treated teethand comprised 78.1% rotary NiTi files, 15.9% SS hand files, 4.0% pastefillers, and 2.0% lateral spreaders. The frequency of rotary NiTi instru-ment breakage was comparable to that previously reported for handfiles.

b) In the hands of experienced operators, endodontic instrumentfracture, in particular rotary NiTi, had no adverse influence on theoutcome of nonsurgical root canal treatment and retreatment when theinstrument remained in the root canal. The presence of a preoperative

TABLE 5. Healing according to recall period (yr) for teeth

No. ofteeth (n) C I U F % healed

(C�I)

Fracturedinstrument, (yrs)1–2 83 56 18 1 8 89.22–4 42 36 3 0 3 92.9�4 21 20 1 0 0 100

146 112 22 1 11 91.8Control, (yrs)

1–2 83 53 26 0 4 95.22–4 45 38 6 0 1 97.8�4 18 14 1 0 3 83.4

146 105 33 0 8 94.5

C � complete healing, I � incomplete healing, U � uncertain, F � no healing.

Clinical Research

JOE — Volume 31, Number 12, December 2005 Fractured Instruments and Treatment Outcome 849

periapical radiolucency, rather than the fractured instrument per se,was a more clinically significant prognostic indicator.

AcknowledgmentsThe authors would like to thank the following for their valuable

assistance in the research project: Associate Professor Ian Gordon ofthe Statistical Consulting Centre, University of Melbourne, Drs. Viv-ian Jeng and Pairoj Linsuwanont of the School of Dental Science,University of Melbourne, and all the endodontists—in particular,the practice principals Drs. Garry Nervo and Peter Parashos— of theparticipating endodontic practices for their cooperation, patience,and contribution of cases to the project.

References1. Strindberg LZ. The dependence of the results of pulp therapy on certain factors: an

analytical study based on radiographic and clinical follow-up examinations. ActaOdontol Scand 1956;14(Suppl 21):1–175.

2. Grossman LI. Guidelines for the prevention of fracture of root canal instruments. OralSurg Oral Med Oral Pathol 1969;28:746 –52.

3. Ingle JI, Glick D. The Washington study. In: Ingle JI, ed. Endodontics, 1st ed. Phila-delphia: Lea & Febiger, 1965:54 –77.

4. Crump MC, Natkin E. Relationship of broken root canal instruments to endodonticcase prognosis: a clinical investigation. J Am Dent Assoc 1970;80:1341–7.

5. Fox J, Moodnik RM, Greenfield E, Atkinson JS. Filling root canals with files. Radio-graphic evaluation of 304 cases. NY State Dent J 1972;38:154 –7.

6. Ward JR, Parashos P, Messer HH. Evaluation of an ultrasonic technique to removefractured rotary nickel-titanium endodontic instruments from root canals: an exper-imental study. J Endod 2003;29:756 – 63.

7. Ward JR, Parashos P, Messer HH. Evaluation of an ultrasonic technique to removefractured rotary nickel-titanium endodontic instruments from root canals: clinicalcases. J Endod 2003;29:764 –7.

8. Ruddle CJ. Nonsurgical retreatment. J Endod 2004;30:827– 45.9. Sackett D, Richardson W, Rosenberg W, Haynes R. Evidence-based medicine: how to

practice and teach EBM. London: Churchill Livingstone, 1997.10. Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies,

and the hierarchy of research designs. New Engl J Med 2000;342:1887–92.11. Bergenholtz G, Lekholm U, Milthon R, Heden G, Odesjo B, Engstrom B. Retreatment

of endodontic fillings. Scand J Dent Res 1979;87:217–24.12. Molyvdas I, Lambrianidis T, Zervas P, Veis A. Clinical study on the prognosis of

endodontic treatment of teeth with broken endodontic instruments. Stoma 1992;20:63 (In Greek). Data cited in: Risk management of root canal treatment. LambrianidisI, ed. Thessaloniki: University Studio Press, 2001: 199 –247.

13. Engstrom B, Hård L, Segerstad AF, Ramstrom G, Frostell G. Correlation of positive

cultures with the prognosis for root canal treatment. Odontol Revy 1964;15:257–70.14. Engstrom B, Lundberg M. The correlation between positive culture and the prognosis

of root canal therapy after pulpectomy. Odontol Revy 1965;16:193–203.15. Kerekes K, Tronstad L. Long-term results of endodontic treatment performed with a

standardized technique. J Endod 1979;5:83–90.16. Sjogren U, Hagglund B, Sundqvist G, Wing K. Factors affecting the long-term results of

endodontic treatment. J Endod 1990;16:498 –504.17. Pettiette MT, Conner D, Trope M. Procedural errors with the use of nickel-titanium

rotary instruments in undergraduate endodontics. J Endod 2002;28: 259 [abstractPR 24].

18. Al-Fouzan KS. Incidence of rotary ProFile instrument fracture and potential for by-passing in vivo. Int Endod J 2003;36:864 –7.

19. Schafer E, Schulz-Bongert U, Tulus G. Comparison of hand stainless steel and nickeltitanium rotary instrumentation: a clinical study. J Endod 2004;30:432–5.

20. Parashos P, Messer HH. Questionnaire survey on the use of rotary nickel-titaniumendodontic instruments by Australian dentists. Int Endod J 2004;37:249 –59.

21. Parashos P, Gordon I, Messer HH. Factors influencing defects of rotary nickel-tita-nium endodontic instruments after clinical use. J Endod 2004;30:722–5.

22. Landis RJ, Koch GG. The measurement of observer agreement for categorical data.Biometrics 1977;33:159 –74.

23. Friedman S. Treatment outcome and prognosis of endodontic therapy. In: Ørstavik D,Pitt Ford TR, eds. Essential endodontology: prevention and treatment of apical peri-odontitis. Oxford: Blackwell Science, 1998:367–91.

24. Torabinejad M, Lemon RR. Procedural accidents. In: Walton RE, Torabinejad M, eds.Principles and practice of endodontics, 3rd ed. Philadelphia: W.B. Saunders Com-pany, 2002:310 –30.

25. Grahnen H, Hansson L. The prognosis of pulp and root canal therapy. Odontol Revy1961;12:146 – 65.

26. Bystrom A, Happonen R-P, Sjogren U, Sundqvist G. Healing of periapical lesions ofpulpless teeth after endodontic treatment with controlled asepsis. Endod Dent Trau-matol 1987;3:58 – 63.

27. Ørstavik D. Time-course and risk analyses of the development and healing of chronicapical periodontitis in man. Int Endod J 1996;29:150 –5.

28. Molven O, Halse A, MacDonald-Jankowski D. Periapical changes following root-canaltreatment observed 20 –27 years postoperatively. Int Endod J 2002;35:784 –90.

29. Ørstavik D, Kerekes K, Eriksen HM. The periapical index: a scoring system for radio-graphic assessment of apical periodontitis. Endod Dent Traumatol 1986;2:20 –34.

30. Trope M, Delano O, Ørstavik D. Endodontic treatment of teeth with apical periodon-titis: single vs. multivisit treatment. J Endod 1999;25:345–50.

31. Cvek M, Granath L, Lundberg M. Failures and healing in endodontically treatednon-vital anterior teeth with posttraumatically reduced pulpal lumen. Acta OdontolScand 1982;40:223– 8.

32. Weiger R, Rosendahl R, Lost C. Ifluence of calcium hydroxide intracanal dressings onthe prognosis of teeth with endodontically induced periapical lesions. Int Endod J2000;33:219 –26.

Clinical Research

850 Spili et al. JOE — Volume 31, Number 12, December 2005